Automating the Design of Game Visualizations

In general game playing [5], game rules are communicated at runtime, and the players must be able to read and understand the descriptions they are given in order to play legally and effectively. Unlike specialized game players, such as Deep Blue, general game players cannot rely on algorithms designed in advance for specific games. General game playing expertise depends on intelligence on the part of the game player and not merely intelligence of the programmer of the game player. In general game playing, games are defined in a formal language known as Game Description Language (GDL), which is similar to the logic programming languages PROLOG and Datalog. At runtime, the GDL description of a game is communicated to the game players. Since general game players are primarily computer programs which are adept at reading GDL description of games and playing them effectively, the general game playing community has not placed much emphasis on the game visualizations. However, recent General Game Playing competitions [4] feature a Carbon versus Silicon contest where a human player plays against the best general game playing program in the competition. Humans, unlike computer programs, need a visual medium to understand, strategize and play games. Hence, human players must be provided with a game visualization in addition to supplying them with the game description in GDL. Currently, the need for a game visualization is being served by game authors having to specify a style sheet [1] along with the game rules. However, the visualization created by an author may not be ideal for a gameplayer. For example, suppose that the game of checkers in its standard visualization with red and black pieces is presented to color-blind people. In this case the game visualization is ineffective since the players cannot distinguish between red and black pieces and cannot, there-